本帖最后由 xzlcqxy 于 2016-3-30 15:19 编辑

%清除变量空间

clc;clear;close all

%%

%数据导入、预处理

t=[0 6 10 11 12 13 14 15 16 17 18 22 26 30 36];

P=t';

a=[0 0 0 0.2 0.34 0.4 0.45 0.5 0.49 0.44 0.4 0 -0.17 0 0];

T=a';

%数据可视化

figure(1);subplot(1,2,1);plot(T);title('原始数据','FontSize',10);grid on

%LS-SVM模型参数初始化

type = 'function estimation';

kernel='RBF_kernel';

preprocess='original';

%参数寻优

gam = 10000;

sig2 = 0.6;

model = initlssvm(P,T,type,gam,sig2,kernel);             % 模型初始化

costfun = 'crossvalidatelssvm';

costfun_args = {10,'mse'};

optfun = 'gridsearch';

model = tunelssvm(model,optfun,costfun,costfun_args);   % 模型参数优化

%进行模型训练

model = trainlssvm(model);

%回归预测

predictlabel = simlssvm(model,P);

figure(1);subplot(1,2,2);plot(predictlabel);title('回归预测数据','FontSize',10);grid on

%%

%测试样本回归

t_test=[11 12 13 14 15 16 17 18 22];

P_test=t_test';

a_test=[0.2 0.34 0.4 0.45 0.5 0.49 0.44 0.4 0];

Y_test=a_test';

figure(2);subplot(1,3,1);plot(Y_test);title('测试样本数据','FontSize',10);grid on

Y=simlssvm(model,P_test);

figure(2);subplot(1,3,2);plot(Y);title('测试样本回归预测数据','FontSize',10);grid on

figure(3);subplot(1,2,1);plot(t,a,'*');

hold on

plot(t,predictlabel,'*','color','r');title('结果对比','FontSize',10);grid on

figure(3);subplot(1,2,2);plot(t,a);

hold on

plot(t,predictlabel,'color','r');title('结果对比','FontSize',10);grid on

%预测11~18

[alpha,b] = trainlssvm({P,T,type,gam,sig2,'RBF_kernel','preprocess'});

t_test=[11 12 13 14 15 16 17 18 22];

P_test=t_test';

a_test=[0.2 0.34 0.4 0.45 0.5 0.49 0.44 0.4 0];

Y_test=a_test';

Y=simlssvm({P,T,type,gam,sig2,'RBF_kernel','preprocess'},{alpha,b},Y_test)

这样就行了~